The adherence-associated Fdp fasciclin I domain protein of the biohydrogen producer Rhodobacter sphaeroides is regulated by the global Prr pathway

(cid:1) Adherence and immobilization of hydrogen-producing microbes improves H 2 yields. (cid:1) fdp encodes an adherence factor in hydrogen-producer Rhodobacter sphaeroides . (cid:1) Expression of fdp is negatively regulated by the global Prr regulatory pathway. (cid:1) Negative regulation by Prr was demonstrated under all growth conditions tested. (cid:1) Fdp-based adherence may be optimised by inactivating the PrrA binding site(s) of fdp. Expression of fdp, encoding a fasciclin I domain protein important for adherence in the hydrogen-producing bacterium Rhodobacter sphaeroides, was investigated under a range of conditions to gain insights into optimization of adherence for immobilization strategies suitable for H 2 production. The fdp promoter was linked to a lacZ reporter and expressed in wild type and in PRRB and PRRA mutant strains of the Prr regulatory pathway. Expression was signiﬁcantly negatively regulated by Prr under all conditions of aerobiosis tested including anaerobic conditions (required for H 2 production), and aerobically regardless of growth phase, growth medium complexity or composition, carbon source, heat and cold shock and dark/light conditions. Negative fdp regulation by Prr was reﬂected in cellular levels of translated Fdp protein. Since Prr is required directly for nitrogenase expression, we propose optimization of Fdp-based adherence in R. sphaeroides for immobilized biohydrogen production by inactivation of the PrrA binding site(s) upstream of fdp .


Introduction
Rhodobacter sphaeroides belongs to the purple non-sulphur (PNS) group of bacteria that are widely recognized as potential 'green energy' producers of biohydrogen from solid food waste and food processing wastewater (reviewed in Refs. [1e4]). There are several other bacterial groups that generate hydrogen such as the bio-photolytic microalgae and cyanobacteria [5,6], and some acidogenic thermophiles and mesophiles that perform dark fermentative hydrogen production [7e9]. Examples recently reported include hydrogenproducing clostridial strains isolated from landfill leachate sludge [10], that produce high yields of up to 4.7 mol H 2 /mol glucose [11]; Clostridium sartagoforme and Enterobacter cloacae strains isolated from Sago industrial effluent [12]; extreme halophiles that produce biohydrogen from lignocellulose biomass in nearly saturated salt [13]; Bacillus spp. isolated from banana waste [14] and improved hydrogen production by bioaugmentation with thermophiles exampled by Thermoanaerobacterium thermosaccharolyticum used to enhance thermophilic hydrogen production from corn stover hydrolysate [15]. The use of consortia of these groups of microorganisms, derived either as endogenous species isolated from biomass or from other environmental sources and used to augment the natural microbial flora has also proved a successful strategy [16e20]. However, the photofermentative processes involved in hydrogen production performed by the PNS group (represented by R. sphaeroides but also including Rhodopseudomonas capsulatus, R. palustris and Rhodospirillum rubrum), has attracted more attention because of the higher conversion efficiency and yields expected from the conversion of substrate to hydrogen and the abilities to utilise food industry wastes and solar light energy of wide ranging wavelengths (522e860 nm) [1,3,21].
R. sphaeroides has attracted particular attention, not least because of its remarkable metabolic versatility; it is able to grow photoheterotrophically, photoautotrophically, fermentatively and using aerobic or anaerobic respiration [21e24]. Photofermentation by PNS bacteria such as R. sphaeroides involves fermentation of organic substrates in the presence of light. Light results in the production and activity of a photosynthetic apparatus which facilitates electron flow from substrate to the [MoeFe]-nitrogenase. Nitrogenase activity results not only in fixation of nitrogen (in an irreversible reaction requiring large amounts of ATP via the F 0 F 1 -ATPase), but also conversion of H þ to hydrogen gas. R. sphaeroides also has a [NieFe]-Hyd uptake hydrogenase enzyme which catalyses H 2 oxidation in the presence of hydrogen gas. Although this enzyme also produces hydrogen under nitrogen excess conditions it is the nitrogenase that is considered to be the most important source of hydrogen generation [4]. Thus, hydrogen production in R. sphaeroides and other PNS bacteria provides a substrate for hydrogen oxidation reactions for energy generation and facilitates the activities of nitrogenase which catalyses the fixation of atmospheric nitrogen into a cellular source of reduced nitrogen [1,2].
There are a number of external factors reported to influence hydrogen production by R. sphaeroides, including culture medium composition (including nitrogen source and concentration, choice of organic substrate, use of mixed carbon sources and incorporation of certain metal ions), reducing agents, pH, light-dark period, illumination intensity, temperature, aerobiosis conditions and even low-intensity electromagnetic fields (e.g. Refs. [25,26] and reviewed in Refs. [2,4,9]). R. sphaeroides has been successfully used for biohydrogen production from biomass; e.g. it has recently been trialled for single-stage hydrogen production from hydrolyzed straw [27] and sugar beet molasses [28], and recently a new strain was identified for producing hydrogen using oil palm waste hydrolysate [29]. It has also been successfully used in co-culture with Enterobacter aerogenes for hydrogen production using Calophyllum inophyllum oil cake as complex carbon source [30]. Therefore, much is known about the external conditions needed to obtain and increase hydrogen production, though not all the mechanisms by which they work are yet understood.
Immobilization of PNS bacteria through biofilm formation has also been reported to be beneficial for hydrogen yields and opens up the possibilities of semi-or full-continuous culture methods for hydrogen production [1,31e33], including biophotoreactor technologies with enlarged surface areas [34e36]. Biofilm formation and adherence properties in R. sphaeroides are multifactorial, affected by flagellar location and number [37,38], chemotaxis [39], membrane cardiolipin [40], presence of functional fasciclin-1 domain protein (Fdp) [41], as well as by light-driven and other regulatory factors [42,43]. In the case of R. sphaeroides Fdp, insertionallyinactivated fdp knockout strains were reported to reduce cell adherence by 100-fold (in terms of cell number) [41]. Fdp resembles the fasciclin I (FAS1) domains found in proteins of higher organisms that have important roles in cell adhesion (Fig. 1). It also shares 60% identity (74% similarity) with the nodule-expressed Nex18 protein of Sinorhizobium meliloti [44], though there appear to be no homologues in other PNS bacteria (Fig. 1). The precise mechanism by which Fdp promotes cell adherence (a prerequisite for biofilms) in R. sphaeroides remains unknown [41]. Clearly, a deeper understanding of the factors important for establishment and maintenance of R. sphaeroides in an immobilized state will be important for improved hydrogen yields reportedly gained through immobilization, not least through employment of continuous flow photobioreactors which optimize microbial exposure to light and fresh nutrients and biomass substrates [e.g. 35]. The aim therefore of the present study was to identify conditions for Fdp expression in R. sphaeroides that promote immobilization and which can therefore ultimately be applied to hydrogen production via nitrogenase. This was investigated by testing a range of growth, chemical and physical conditions on transcriptional expression of fdp, including anaerobic conditions with reduced NH 4 þ . We show that fdp transcription is strongly repressed by the Prr global regulatory system in wild type R.
sphaeroides under all laboratory conditions tested here. This leads us to propose the future development of a new strain mutagenesis strategy for optimizing hydrogen generation based on increased attachment and biofilm development mediated by Fdp in R. sphaeroides for use in bioreactors designed for continuous biohydrogen production, through promoter engineering upstream of the fdp gene that reduces or abolishes Prr repressor binding upstream of the fdp locus.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( x x x x ) x x x

Chemicals
The chemicals used in this study were purchased from Merck (Gillingham, Dorset, UK), VWR (Lutterworth, Leicestershire, UK), or Fisher Scientific (Loughborough, Leicestershire, UK) unless otherwise stated, and were of molecular biology grade.

Bacterial strains, plasmids and growth conditions
All strains and plasmids are described in Table 1 [45e49]. E. coli strains DH5a, S17-1 and BL21[DE3] have been described previously and were routinely cultured aerobically in Luria-Bertani (LB) media by vigorous aeration of culture vessels, or on LB agar, at 37 C as described in Ref. [50]. Where appropriate, media were supplemented with 50 mg mL À1 ampicillin and/or 50 mg mL À1 kanamycin or 500 mg mL À1 carbenicillin. Reporter plasmid transfer into R. sphaeroides was by conjugative transfer from E. coli S17-1 [45]. R. sphaeroides NCIB 8253 was cultured at 34 C in M22 medium [45]; the fdp and prrA mutant strains were cultured in M22 containing 20 mg mL À1 kanamycin. Liquid M22 lacked added caesamino acids and contained 1.5 mM NH 4 þ which permits some nitrogenase expression under anaerobic conditions [51] but little/no hydrogen evolution anaerobically due to the absence of light and the presence of dissolved N 2 . Growth was measured using culture absorbance at 680 nm (A 680 ). Aerobic growth of R. sphaeroides was achieved using vigorous shaking of 10 mL medium in 250 mL vessels or 500 mL in 2 L vessels. Semi-aerobic growth at 34 C was carried out using 70 mL medium in 250 mL vessels, whilst anaerobic growth at 34 C in the dark was achieved using M22 medium containing 60 mM dimethyl sulphoxide (DMSO). R. sphaeroides prrA and prrB knockout mutants (PRRA and PRRB respectively) were constructed by transposon Tn5 mutagenesis of pREG464 [52,53]. Insertion sites in prrA or prrB were verified by restriction analysis and DNA sequencing. Kanamycin-resistant transconjugants were screened for loss of the suicide plasmid by Southern hybridization using parental pSUP202 as labeled probe. The correct location of the inserted transposon (and loss of intact prrA or prrB gene from the chromosome) was determined by restriction and Southern hybridization analysis. The phenotypes of the resulting strains were identical to those reported for these mutations previously [54,55], including photosynthesis-and nitrogenase-minus phenotypes, and were successfully complemented using a 4.8-kb BamHI prr (reg) fragment described in Ref. [49].
Plasmids pBluescript-SK and pET14b have been described previously [53]. Plasmid pSUP202 is a R. sphaeroides suicide plasmid used in fdp and prr mutant construction and is the host plasmid for the R. sphaeroides genomic library and has been described previously [47]. The R. sphaeroides replicative pSDP1 reporter plasmid possessing a promoter-less lacZ gene, and pUX-Km, have both been described previously [48].
Isolation of the fdp gene has been described previously [56]. Construction of an insertionally-inactivated fdp mutant was described by Ref. [41].

Reporter studies of fdp expression
The 592 bp promoter region of the fdp gene (À613 to À21 relative to the ATG start codon) was amplified by polymerase i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( x x x x ) x x x chain reaction (PCR) using pSUP202fdp-13 as template, and primers NIT1: 5 0 -ATGAGGTACCTCGAGGAGGGTCCG-CAGCTCG -3 0 and NIT2: 5 0 -ATGATCTA-GATCGTGCCGTGCTTGGCCTGC -3 0 (KpnI and XbaI sites are underlined). The PCR fragment was purified and cloned into SmaI-cut pBluescript-SK to give pBSFDPP-2. The promoter insert was checked by sequencing, prior to digestion with KpnI and XbaI, and ligation into KpnI-and XbaI-digested pSDP1, a promoter-less lacZ reporter plasmid described previously [48]. The resulting plasmid, pSDP-FDPP, was checked by restriction analysis. pSDP-FDPP and control pSDP1 were introduced into R. sphaeroides wild type and PRR mutant strains by conjugation via E. coli S17-1. Transconjugants were selected in culture media containing 1 mg mL À1 tetracycline (plus 20 mg mL À1 kanamycin for the PRR strains). b-galactosidase reporter assays were performed as described [48]. Protein content was measured either using the Bio-Rad DC Protein Assay or by the method in Ref. [57]. Bovine serum albumin (Sigma-Aldrich, Poole, UK) was used as the calibrant.

Separation of cell proteins by two-dimensional SDS-PAGE and identification of Fdp
Protein extracts of semi-aerobically grown R. sphaeroides were prepared by batch culture to mid-exponential phase (A 680 of 0.6). Cells from 300 ml cultures were harvested by centrifugation at 4 C and resuspended in 10 mL TGEND buffer (comprising 10 mM Tris.HCl pH 8.0, 10% (v/v) glycerol, 0.1 mM EDTA, 50 mM NaCl, 0.1 mM dithiothreitol (DTT), 500 mM phenylmethyl sulfonyl fluoride (PMSF) and 50 mM N-tosyl-Lphenylalanine chloromethyl ketone (TPACK); final pH 8.3) at 4 C. Cell suspensions were sonicated on ice (4 Â 15 s bursts with 45 s intervals on ice). Unbroken cells and cell debris were removed by centrifugation for 20 min at 29,000 g at 4 C and the protein supernatents (soluble, cytoplasm plus periplasm) stored at À70 C.
40 mg protein were diluted in rehydration buffer (8 M urea, 2% Triton X-100, Pharmalyte pH3-10, Amphiline pH 6e8 1.5%, DTT 100 mM, bromophenol blue trace) and applied to 11 cm Immobiline DryStrips (Pharmacia Biotech Inc, USA) with an immobilized pH nonlinear gradient, pH 3 to 10. The first dimension was performed on an IGP isoelectric focusing unit (Pharmacia Biotech Inc, USA), and the second dimension was performed in 8e18% polyacrylamide gels. Gels were stained either with Coomassie brilliant blue or by the modified silver staining method of [58]. For sequence determinations of native Fdp, proteins were blotted onto membrane, visualised with Coomassie Brilliant Blue stain, excised from the membrane and the N-terminal sequence determined by Edman degradation.
Overexpression and purification of His 6 -tagged Fdp in E. coli

BL21[DE3]
To overexpress Fdp, the fdp region 57 to 470 (relative to ATG, where A is position 1), which lacks the region encoding the signal peptide region (residues 1e18), was amplified by polymerase chain reaction using upstream primer SGINT1: 5 0 -TCAGCCATATGGAAACCGGAGACATCGTGGA -3 0 (NdeI cloning site underlined), and downstream primer SGEL2: 5 0 -GCTAG-GATCCGCATCAGGCGCCCGGCATCAGCACG -3 0 (BamHI site underlined), using pSUP202fdp-13 as template. The 470-bp fragment was purified by gel extraction and cloned into SmaI-digested pBluescript-SK to give pBlFDP470. The presence of inserts with correct sequence was verified by restriction digest analysis and sequencing. Plasmid pBlFDP470 was digested with BamHI and NdeI, and the fdp fragment cloned into pET14b (Novagen® Merck Group, UK). The final expression construct, pETfdp470, expresses a Fdp protein with a Nterminal MGSS(H) 6 SSGLVPRGSHM sequence followed by Fdp starting at E-19. Verification of the N-terminal sequence of recombinant purified Fdp was performed by Edman degradation: approximately 3 mg of purified his-tagged Fdp was loaded onto 15% polyacrylamide resolving gels, and transferred to Fluorotrans™ membrane (Pall BioSupport, UK) by electroblotting for 1 h at 100 V using a Bio-Rad Mini Trans-Blot Cell. The proteins were visualised with Coomassie Brilliant Blue, excised from the membrane and the N-terminal sequence determined.

Western blotting
To verify the presence of recombinant His-tagged Fdp purified from IPTG-induced E. coli BL21[DE3]/pET14fdp, Western blotting was undertaken using an antibody that recognises the His 6 motif as described previously [59]. Briefly, purified His 6 -Fdp (4 mg) was loaded on 15% SDS-polyacrylamide resolving gels. Following electrophoresis by standard methods [50], proteins were transferred to nitrocellulose membrane (Amersham Hybond-C) by electroblotting for 1 h at 100 V using a Bio-Rad Mini Trans-Blot Cell. The transfer buffer contained 25 mM Tris.HCl pH 8.3, 192 mM glycine, 20% methanol, 0.025% sodium dodecyl sulphate (SDS). Membranes were washed twice for 10 min with TBS buffer (10 mM Tris.HCl pH 7.5, 150 mM NaCl) at room temperature, and incubated for 16 h in 3% (w/v) bovine serum albumin in TBS buffer. Membranes were then washed twice for 10 min each time in TBSTT buffer (TBS buffer containing 0.05% (v/v) Tween-20, 0.2% (v/v) Triton X-100), and then once for 10 min in TBS buffer. A 1:1000 dilution of mouse anti-RGS(H) 6 monoclonal antibody (Qiagen Ltd, Manchester, UK) was then prepared in TBS containing 3% BSA into which membranes were immersed for 1 h at room temperature. Following two washes for 10 min each time in TBSTT buffer and one wash for 10 min in TBS at room temperature, a 1:5000 dilution of goat anti-mouse IgG horse radish peroxidase conjugate (Stratech Scientific Ltd, Ely, UK) in TBS containing 10% (w/v) skimmed milk powder was added and the membranes incubated for 1 h at room temperature. Following four washes for 10 min each in TBSTT buffer, i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( x x x x ) x x x membranes were incubated with ECL Western blotting detection reagent (GE Healthcare, USA) and developed by autoradiography with Xograph film (Kodak Co., Herts, UK).

Mass determinations using electrospray mass spectrometry
Samples of purified recombinant Fdp were prepared for electrospray mass spectroscopy by the method of [60] and analysed on a single quadrupole, bench top mass spectrometer (Platform II, Micromass UK Ltd). as described by Ref. [53]. Samples were dissolved in formic acid:methanol:water (1:1:1, v/v/v) and infused into the ionisation source at a flow rate of 10 mL per minute. Data were acquired over the appropriate m/z range and were processed using the MassLynx software supplied with the instrument. The m/z spectrum was transposed onto a true molecular mass scale for more facile identification using Maximum Entropy processing techniques. An external calibration is applied, using horse heart myoglobin (MW 16,951.49 Da) as the calibrant.

Protein determinations
Protein content was measured using the Bio-rad DC Protein Assay Kit II (Bio-rad Laboratories Inc., Watford, Herts., UK) as outlined by the manufacturer, using bovine serum albumin as the standard.

Confirmation of Fdp as a member of the fasciclin I protein superfamily
The open reading frame encoding Fdp (ORF RSP1409) was first identified as a FAS1 fasciclin Ielike protein in Ref. [56] (Beta-Ig-H3/Fasciclin; https://www.uniprot.org/uniprot/Q3IXZ6). Fdp has a predicted signal peptide at the N-terminus (residues 1e18: (RKTLLALSLGLLAAPAFA)) suggesting a protein that is translocated across the inner membrane resulting in a mature 137-residue protein possessing the N-terminal sequence ETGDIVETATGA. By PSI-BLAST, the closest sequence similarity (60% identical; 74% similar) is to Sinorhizobium meliloti Nex18 (Fig. 1). It is also related (32e39% identity; 52e59% similarity) to Mycobacterium tuberculosis MPT70 and M. bovis MPB70 major secreted proteins [61], the fasciclin I domains of mammalian transforming growth factor b-induced proteins (bIG-H3 or RGD-CAP adhesion proteins, as indicated in Uni-Prot) [62], and human osteoblast-specific factor 2 (OSF-2 or periostin) [63], which is thought to be involved in bone adhesion and is a ligand for avb5 integrin [64] (Fig. 1). Drosophila FAS1 domain 4 [65,66], which is responsible for axon guidance, has 29% identity to Fdp (Fig. 1). The common feature in all these proteins, where a function is known, is their involvement in protein-protein associations. The sequence similarities are quite striking, since fasciclin I domains generally exhibit low overall sequence conservation (<20%) [66]. The two regions of high conservation recognized for the FAS1 superfamily (H1 and H2) are also strongly conserved in this putative protein. Taken together with the NMR structure of Fdp described previously [41,56], it is clear that this protein is a member of the fasciclin I protein superfamily. One unusual aspect of this particular fasciclin-domain protein is that it occurs in a free-living bacterium, and fortuitously this freeliving species is well characterized regarding its physiology, metabolic versatility, molecular bases for responses to environmental change and it is also amenable to knock-out strategies. Indeed the role of Fdp in cell adherence properties of R. sphaeroides has already been established; Fdp appears to promote cell adherence as shown by insertional activation studies in which inactivation of fdp resulted in a 100-fold reduction in numbers of adherent cells in a R. sphaeroides adherence assay [41]. Here we investigate the regulation of expression of this adherence factor in R. sphaeroides, which could yield important knowledge for the establishment and continuous immobilization of bacterial cells in bioreactors.

Transcription of Fdp is negatively regulated by the Prr signaling pathway under anaerobic and other growth conditions
Prr is a major regulator that senses changes in external redox potential and serves as a global switch in gene expression for many genes in R. sphaeroides [67e69]. To investigate whether this global environment-responsive regulator controls fdp transcription, reporter studies were undertaken using the promoter region of the fdp gene linked to a lacZ reporter gene, which was expressed in both wild type and PRR mutants. Table 2 shows activity of the fdp promoter under different aerobiosis conditions, as shown by b-galactosidase measurements of R. sphaeroides extracts from stationary-phase cells harbouring pSDP-FDPP, a pRK415-based replicative reporter plasmid carrying 592-bp of fdp upstream sequence transcriptionally linked to lacZ. Experiments were carried out using wild type, plus two mutants PRRA and PRRB in which the prrA (encoding the response regulator PrrA) and prrB (encoding the redox sensor kinase PrrB) genes, respectively, were insertionally inactivated. Anaerobic conditions were achieved using dark conditions in the presence of DMSO rather than light conditions for light harvesting, since PRR mutants are unable to grow photosynthetically. Table 2 shows that levels of fdp expression levels in aerobic and semi-aerobic cells of wild type grown on succinate-lactate medium were similar (DA 405 units/min/mg protein ¼ 93e100 Â 10 3 ), but were slightly lower under anaerobic conditions (required for nitrogenase expression and thereby hydrogen generation) [1] (and under which the Prr pathway generates a higher level of phosphorylated PrrA, Prr-P) (DA 405 units/min/mg protein ¼ 64 Â 10 3 ) [70] ( Table 2). Expression was significantly higher (3.7e40.3-fold) in both PRRA and PRRB strains compared with wild type under all conditions of aerobiosis in these cells grown on glucose or succinate-lactate (Table 2), demonstrating that the Prr system exerts negative control of fdp transcription under both aerobic and anaerobic nitrogenase-expressing conditions. Presumably sufficient transcriptionally-active PrrA or PrrA-P must occur for the efficient repression of the fdp promoter region observed under all aerobiosis conditions. The fold effect on expression levels in PRR mutants appears to be less marked under increasingly anaerobic conditions (though nonetheless significant), possibly suggesting that PrrA (which predominates under aerobic conditions compared with PrrA-P), is the overall repressor, and/or alternatively that additional aerobiosis-responsive regulators are regulating to different degrees under these conditions.
Expression of fdp was less elevated in the PRRB strain compared with PRRA under aerobic and semi-aerobic growth conditions of aerobiosis, but levels were approximately equivalent in PRRA and PRRB strains under anaerobic conditions. This suggests that whilst there is a role for PrrB in fdp regulation under all aerobiosis conditions (shown by the elevated levels of reporter in the PRRB strain), under anaerobic conditions the loss of PrrB in PRRB exerts no greater or lesser effect on fdp transcription than loss of PrrA-P in PRRA, suggesting that PrrA-P derived only from PrrB acts as the repressor under anaerobic conditions in wild type cells and/or that any additional regulators present exert their effects equally on fdp expression in anaerobically-cultured PRRB and PRRA strains ( Table 2).
A similar trend was observed using glucose-containing medium, though reporter levels (and fold effects) were overall higher in aerobic and semi-aerobic mutant cells compared with those grown in the same aerobiosis conditions using succinate-lactate medium ( Table 2). Under anaerobic conditions on glucose (in common with succinate-lactate), fdp expression levels were elevated 5.0e6.8 fold in the absence of a functioning Prr pathway.
Reporter studies of cells harvested at different times during batch growth revealed that in the wild type strain, under all conditions of aerobiosis including anaerobic conditions, reporter levels remained at constant low levels throughout growth (Fig. 2). By contrast, reporter levels in anaerobic/darkgrown PRRA were significantly elevated, though once again relatively similar throughout growth. Under aerobic/dark conditions (and to a lesser extent under semi-aerobic/dark conditions), reporter levels appeared more variable and possibly growth-phase dependent in the PRRA mutant. Levels in the PRRA mutant increased during lag and early exponential phase under aerobic conditions and reached a maximum level in late-exponential phase, reaching up to 99-fold those of wild type cells in the same phase of growth (Fig. 2). This may suggest the presence of additional regulators governing fdp expression, in addition to Prr. Indeed, the higher fdp expression observed in late exponential phase cells cultivated under aerobic conditions is reminiscent of gene expression control governed by quorum-based systems [71]. To perform preliminary investigations on whether quorum sensing in R. sphaeroides [72] could possibly play a role in regulation of fdp, reporter studies were undertaken using early-exponential phase aerobically-grown cells from wild type and PRRA strains and to which were added sterile culture supernatants from stationary phase wild type cells (shown to accumulate 7,8-cis-N-(tetradecenoyl) homoserine lactone, [72]) to constitute 10% of the total culture volume. Expression levels of fdp were compared to those of untreated cells after 1 h further incubation. Addition of the culture supernatant did not significantly affect expression levels; expression levels in wild type were 0.6-fold compared with untreated cells whilst in the PRRA strain levels were only 1.3-fold those of the control (Table 3). Therefore, these preliminary experiments indicate that quorum sensing plays no detectable role in fdp regulation, but further investigations should be conducted to confirm this.
Although levels of fdp expression were consistently low in wild type (compared with PRR mutants) under all conditions of aerobiosis tested (Table 2), some variation in expression levels nonetheless occurred, specifically there are significantly lower levels of expression under anaerobic conditions on succinate/lactate medium compared with aerobic and semi-aerobic conditions in the same medium (Table 2). To investigate whether other environmental factors can also affect fdp transcription in wild type, the effects of complex versus defined medium, heat versus cold shock and light versus dark conditions were investigated. For comparative purposes, strains were all cultured under aerobic conditions, so that a wider range of conditions could be investigated at a practical level. Thus, the anaerobic conditions required for nitrogenase expression were not specifically investigated  here. The study also included the effects of these factors on fdp expression in PRRA strain, to determine whether any variation also occurs in the absence of the Prr pathway. The results in Table 3 demonstrate that fdp expression was sensitive to growth medium composition in both wild type and PRRA; expression was upregulated 3-fold in the wild type and 4.2-fold in PRRA in Luria-Bertani complex medium compared with M22 succinate-lactate medium (Table 3). There was no significant effect of cold shock in both strains nor of heat shock or light compared with dark on wild type expression. Expression in PRRA was 2.7-fold elevated upon heat shock and 3-fold elevated under light conditions compared with identical dark conditions (Table 3). Thus, growth medium composition (complex versus defined) significantly affected expression in the wild type as well as in PRRA, though most changes in expression levels were observed in the PRRA mutant strain. It is difficult to draw conclusions about the nature of the mechanisms by which such different regulation occurs.
To investigate possible regulatory mechanisms by which fdp is regulated under different growth conditions, studies were focused on reporter studies using pyruvate-grown cells to ascertain whether the global signaling molecule acetyl phosphate which occurs in R. sphaeroides [73] may affect fdp expression. When pyruvate is the carbon source, levels of the small phospho donor acetyl phosphate are elevated [74,75]. Acetyl phosphate is a global signaling molecule that regulates many bacterial cellular processes including nitrogen assimilation, osmoregulation, flagellar biogenesis, pilus assembly, capsule biosynthesis, biofilm development, and pathogenicity   a The fdp promoter region was inserted upstream of lacZ as described in Methods, resulting in pSDP-FDPP (Table 1). b-galactosidase measurements are means derived from triplicate measurements (standard deviation values in parentheses). The enzyme levels produced in corresponding pSDP1-harbouring control cells are shown. b Fold is the ratio of expression levels in the presence and absence of treatment (or Prr mutant: wildtype levels), calculated after subtraction of control pSDP1 activity. See individual treatments for more specific detail. Treatment 1: Activity in early exponential phase cells in response to addition of sterile supernatent from stationary phase cells. c Aerobic dark growth at 34 C (A 680 ) was monitored and either sterile culture supernatent from stationary phase cells were added or no addition made (Control), to early log phase cells. Treatment 2: Activity in stationary phase cells cultured aerobically in rich (LB) and minimal M22 succinate-lactate media. d Aerobic dark growth at 34 C (A 680 ) was monitored (A 680 ) and cells harvested in stationary phase (30e36 h). Fold: ratio of expression levels in LB medium compared with those in M22 medium, calculated after subtraction of control pSDP1 activity. Treatment 3: Activity in stationary phase cells following exposure to temperature shock. e cultures were grown aerobically in M22 succinate-lactate medium at 34 C in the dark until early stationary phase before heat shock at 42 C or cold shock at 5 C for 4 h and cell harvesting. Fold: Ratio of expression levels compared with continued standard conditions, calculated after subtraction of control pSDP1 activity. Treatment 4: Activity in stationary phase cells cultured in the light or dark. f Aerobic growth in M22 succinate-lactate medium at 34 C in the dark or light was monitored (A 680 ) and cells harvested in stationary phase (30e36 h). Fold: ratio of expression levels in the light compared with those in the same medium in the dark under aerobic conditions at 34 C, calculated after subtraction of control pSDP1 activity. g 86 W m 2 .
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( x x x x ) x x x [76]. One way it has been shown to exert its regulatory effects is by direct phosphorylation of response regulators of twocomponent systems, including R. sphaeroides response regulators [74,77,78]. In wild type and fdp mutant strains, pyruvategrown cells consistently expressed significantly less fdp compared with succinate/lactate grown cells, possibly indicating greater repression by Prr, though effects due to additional regulators cannot be ruled out (Table 4). However, in the PRRA mutant strain which lacks the PrrA response regulator, the fold increase in expression levels was significantly higher (83-fold and 58-fold in semi-aerobic and anaerobic conditions, respectively) in pyruvate-grown cells compared with succinate/lactate-grown cells (10-fold) ( Table 4), suggesting a possible role for acetyl phosphate and/or the presence of additional phosphorylatable regulators of fdp expression in addition to the Prr pathway under anaerobic (and semiaerobic) conditions.

Comparisons of Fdp protein levels in wild type and Prr mutants in vivo
As shown in all the reporter data described above, Fdp expression is significantly lower in the wild type strain compared with the PRR mutants. To determine whether these differences are also reflected in vivo with regard to the final translated Fdp protein, 2D SDS-PAGE analysis of cell extracts was undertaken. The Fdp protein possesses a putative signal peptide at the N-terminus, suggesting that the protein is secreted either externally or onto the cell surface, into the periplasm or is membraneeassociated. Since posttranslationally modified Fdp is of relatively low molecular The fdp promoter region was inserted upstream of lacZ as described in Methods, resulting in pSDP-FDPP. All growth experiments were conducted in M22 medium at 34 C in the dark, and anaerobic growth was achieved by supplementing M22 medium with 60 mM dimethyl sulphoxide (DMSO). Growth (A 680 ) was monitored and cells harvested in late exponential phase. b-galactosidase measurements are means derived from triplicate measurements (standard deviation values in parentheses). a Ratio of expression levels compared with wild type, calculated after subtraction of control pSDP1 activity. Soluble extracts (including periplasmic fractions) of washed semi-aerobic wild type and PRR mutant cells were used in the 2D SDS-PAGE analysis (Fig. 3). The correct protein spot and position in the gels were identified by N-terminal sequencing (sequenced as ETGDIVETATSA, compared to the Fdp protein in the R. sphaeroides genome database which is ETGDIVETATGA). Characteristically, it runs anomalously in the approximate technique of SDS-PAGE, with an apparent molecular mass of 17,300 Da, higher than the predicted 13,800 Da. This is a characteristic also observed using a purified his-tagged version of Fdp expressed in E. coli; His 6 -Fdp possesses an apparent molecular mass of 20,100 Da in SDS-PAGE ( Fig. 4) but mass spectrometry reveals a mass of 16003.9 ± 1.6 Da, in good agreement with the expected theoretical value for the recombinant protein (16,004.8 Da). Fig. 3 shows that in PRRA and PRRB soluble extracts, levels of Fdp are significant in comparison with other cellular soluble proteins, confirming that washed cells possess abundant levels of mature, post-translationally modified Fdp. Taken together with the N-terminal sequencing data, these results demonstrate that the predicted N-terminal signal peptide is indeed cleaved in vivo, and that mature Fdp is therefore presumably exported across the inner membrane and at least into the periplasm in these strains. Fdp levels were lower or barely detectable in the wild type strain (Fig. 3), a feature consistent with the findings described above for Fdp transcription.

Discussion
The present study clearly demonstrates that expression of the fdp gene encoding a protein involved in adherence [41] is negatively regulated by the Prr global regulator in R. sphaeroides. Under a wide range of growth conditions tested here including different carbon sources, conditions of aerobiosis (including anaerobic conditions suitable for nitrogenase expression), rich versus defined growth media, heat/cold shock, and light versus dark conditions, elevated fdp promoter activity was consistently observed in PRRA and PRRB mutants compared with wild type, ranging from 3.7 to 154-fold (Tables 2e4). Increased levels of promoter activity in PRR mutants were also observed throughout batch growth under all aerobiosis conditions (Fig. 2) and whilst the highest levels of promoter activity were seen in late exponential phase cells under aerobic/dark conditions, no evidence of quorum-based regulation was found (Table 3), though further study is needed to confirm this. Interestingly Prr repression occurred under all conditions of aerobiosis, suggesting either sufficient levels of PrrA-P under all these aerobiosis conditions for repression, or that unphosphorylated PrrA is also able to repress fdp. The regulatory activity of PrrA (or analogous RegA in other species) as well as of PrrA-P (RegA-P) has been documented previously [77,79e81].
In the absence of the Prr pathway, the increased levels of fdp expression varied depending on growth conditions, possibly suggesting the involvement of additional regulators involved in Fdp regulation. This was further supported by growth experiments using pyruvate as carbon source, in which elevated levels of the global signaling molecule acetyl phosphate are present; in PRR mutants growing on pyruvate, reporter levels were significantly elevated still further (Table  4), suggesting possible regulation by phosphorylatable control systems, such as two-component signal transduction systems. The effects on fdp expression measured using our reporter assay system were also reflected in the levels of translated Fdp protein observed in cell extracts in vivo (Fig. 3).
Evidence for the adherent properties of the wild type strain of R. sphaeroides used in this study has been reported previously [41]. The strain was shown here to exhibit low levels of fdp expression resulting in low, barely detectable levels of post-translationally modified Fdp protein in which the signal peptide has been removed in vivo (Tables 2e4, Fig. 3). Such low levels are surprising, but presumably these levels in the wild type are nonetheless sufficient to support cell adherence. There were low but detectable levels of fdp expression in the wild type under all conditions tested, and yet there were some limited levels of variation in these expression levels under different conditions. For example, expression levels in wild type cells cultured anaerobically in M22 succinate/lactate medium were 64% of those measured aerobically in the same medium (Table 2). Similarly, fdp expression in wild type cells cultured anaerobically with pyruvate as carbon source was approximately 5-fold lower than in cells grown in the same medium semi-aerobically with succinate-lactate as carbon source (Table 4), and aerobic wild type cells cultured in rich LB medium exhibited 3-fold elevated levels of fdp expression over cells cultured in defined M22 medium (Table 3). These variations may be due to variable regulation by Prr itself as reported for the analogous Reg pathway in R. capsulatus which regulates in a variable way different gene sets depending on growth conditions [81]. Alternatively, other Prr-independent regulatory activity may be occurring. In light of the strong regulation exerted by Prr regulation in wild type demonstrated in this study by the significantly derepressed levels of fdp expression observed in the PRR mutants, the latter of these two possibilities appears to be the most likely explanation for the relatively low (but significant) levels of variation seen in the wild type.
The question therefore is why Fdp expression should be subject to such strong regulation by Prr and possibly other phosphorylatable regulators under most growth conditions as demonstrated here. One possible explanation is that there may be occasions in which it is advantageous for the bacterium to experience a full reversal or partial loss of adherence ability, for example in order to enter a motile phase. Perhaps there are particular environmental conditions which occur in nature (and which were not possible to replicate in the laboratory environment), that facilitate full repression of Fdp levels equivalent to a full shut down of Fdp in the wild type, resulting in motile phase non-adherent cells. A previous study established that, in terms of cell numbers, adherence is reduced approximately 100-fold in an insertionally-inactivated fdp mutant [41]. In this regard it is relevant to note that Rhodobacter mutants in the global Prr/Reg regulatory pathway, and shown here to exhibit significantly elevated levels of Fdp in R. sphaeroides, are defective in aerotaxis and motility [81,82]. It is not suggested here that there must therefore be a direct link between elevated Fdp levels and loss in motility and aerotaxis, as Prr is a global regulator affecting many processes, but rather that such characteristics of Prr mutants makes the above hypothesis difficult to test. Another difficulty with testing this possibility is that in the present study no laboratory conditions were identified in which Fdp levels were fully repressed in the wild type. Nonetheless, with regard to promoting permenant adherence and thereby immobilization in a bioreactor environment, we propose that engineering of the PrrA binding site upstream of the fdp promoter to inhibit binding by the PrrA/ PrrA-P repressor may be a useful future strategy. Development of the Prr mutants themselves, which are already lacking PrrA/ PrrA-P binding and produce desirably elevated levels of Fdp, are not suitable in this case as they are defective in expression of nitrogenase for H 2 production and other key metabolic processes such as photosynthesis and CO 2 fixation required under light anaerobic conditions [67,70,81]. Therefore, a mutagenesis strategy designed to specifically abolish or reduce PrrA binding in the fdp promoter region and thereby ensure either strongly elevated levels of Fdp, or levels that are moderately higher than wild type levels, may be a fruitful line of future investigation.
Progress has previously been made in improving hydrogen yields through mutant analysis and genetic/metabolic engineering strategies, mainly targeting the activities of the uptake hydrogenase, poly-3-hydroxybutyric acid synthesis, nitrogenase and light harvesting systems under defined external conditions [2,3,23,83e87]. Not many reports have yet appeared on mutations in transcriptional regulators; however, the studies of [2,87,88] investigated the effects of HupR, HupT, NifA and NifL mutations for improving hydrogen yield in PNS bacteria, with success in improving hydrogen production. Searches using the consensus sequence for PrrA DNA binding in R. sphaeroides [89,90] reveal two possible binding sites for PrrA to the fdp upstream region, both of which occur in the promoter fragment used in the reporter studies described above. One starts at position À432 (5 0 -GCGCCGGCATTCTGCGC). In common with sites of other repressed genes such as hydrogenase (hup), this site has a rather long half-site spacing [89]. The second possible site starts at À419 (5 0 -GCGCCGGATCGC) and possesses a relatively short 6-nt half-site spacing [89]. Following confirmation of these PrrA binding sites, a comprehensive mutagenesis programme can be initiated.

Conclusions
Expression of the fdp gene, which encodes an adherence factor in R. sphaeroides, is negatively regulated by the global Prr regulatory pathway. Strains defective in either the sensor kinase PrrB or the response regulator PrrA of this pathway possess significantly elevated levels of fdp promoter activity, which is also reflected in the levels of translated Fdp protein in R. sphaeroides cells in vivo. One strategy to optimize or increase adherence properties of R. sphaeroides in immobilized bioreactor applications might be to generate altered strains in which the Prr repressor activity has been reduced or removed. Mutations in the prrA or prrB genes themselves is not feasible, as they are required for expression of nitrogenase. We therefore propose targeted mutagenesis of the PrrA binding site upstream of the fdp gene to reduce or remove binding by the PrrA repressor specifically at this site and thereby enhance expression of the fdp adherence factor.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.